Wow... another exceptional post, Ammon. I love how well-researched and balanced your framing is around these complex issues. Your insight that we can't simply opt out, but must instead lean into the tech stack to build systems that prioritize human interests, really resonates.
My mind goes towards immigrants arriving at Ellis Island a century ago - confronted with an overwhelming new world of unfamiliar rules, systems, and power structures. Today, we're all immigrants in this emerging AI-mediated landscape, trying to navigate an environment where the "rules" of engagement are increasingly opaque and potentially adversarial. Anyone interested in buying the Brooklyn Bridge with me? I know a guy that is offering me an exclusive, inside deal... and we can pay with Klarna over a few payments.
The parallel extends further - just as those immigrants needed to develop new frameworks to thrive in their new world, we need new frameworks for digital well-being. We've moved beyond simple heuristics like "buy organic" for healthy living. The challenge is that there's no "certified organic" label for digital content, and the damage from manipulative content can far exceed that of a few pesticide-sprayed grapes.
I'm particularly intrigued by your concept of guardian angel agents. It reminds me of how religious institutions once provided moral guidance through periods of massive social change. Perhaps AI tools, properly designed and aligned with human values, could serve a similar role - helping us maintain our agency while navigating this new landscape. While we're at it, how about some AI generated hymns to go along with it?
The key seems to be combining:
Education about how AI tools actually work and influence us
Critical examination of our personal values and the legacy we want to build
Thoughtful deployment of AI tools that protect rather than exploit
Thank you as always for helping to fire the neurons!
Love your thoughts here Stephen! I'm feeling more and more motivated to work on the guardian angel. It seems the headwinds of unregulated tech and a bifurcation of trust will make adoption tricky — not the mention the impending arms race of AI agent persuasion. But damned if it's not an exciting mission!
Good post man. The social component can't be overstated. Social media and platforms are a means by which individuals obtain presence, affirmation, acknowledgement, and other passive and active forms of social validation. I think this is true for people who are mainly "lurkers" as well as those who are "influencers." Fact that these social relationships are in part imaginary, or at least projected and internalized, doesn't seem to mitigate their power in a society within which so much communication and social interaction is mediated. Mediated relationships have become as "real" as real life, and we practice the same types of social co-presence online as we would in real face to face encounters (albeit without same consequences). So the impact of social platforms is social/relational/psychological as it is also communication/distribution/content. Two axes by which to critique this mode of social interaction and production of social narratives.
The AI angle is interesting and I've thought a fair bit about it. To date AI has not leveraged social sharing; we use GPT privately (or for/at work). LLMs have a tough time with sentiment analysis, personalization, individual preference mapping, etc, so while interesting it's not there yet either as a mode of integrating AI into social platforms or for the guardian angel role you describe. (Social platforms e.g. Facebook might be rolling out AI avatars but think this is just proof of concept stuff as personalization and emotional realism are difficult frontiers for models trained on aggregated/generic linguistic content.) We should chat on LI about this - lack of social sharing w GPT, Claude, Gemini is worth fleshing out - what would they have to change to become acceptable sources, or for interactions with LLMs to be socially valid conversations?
Thanks for the comments Adrian! Thoughtful as usual :)
re. mediated relationships... I often think about Dunbars Number, a pre-digital take on our capacity for social connection. One way to think of the impact of social media is it's robbing our potential capacity for nurturing our community relationships. Virtual relationships are empty calories.
re. social + AI... it's coming. Meta is sitting on a massive pile of personalization insights, but have been constrained by the dynamics of the ad model, plus the meager regulatory constraints around the ad market. Embedding AI into the marketing tools to contextually deliver persuasive content will fly under the radar of current regulation. It's the low hanging fruit to boost their ad business and drive a much higher hit rate. Lots they can do without directly connecting AI to consumers.
Wow... another exceptional post, Ammon. I love how well-researched and balanced your framing is around these complex issues. Your insight that we can't simply opt out, but must instead lean into the tech stack to build systems that prioritize human interests, really resonates.
My mind goes towards immigrants arriving at Ellis Island a century ago - confronted with an overwhelming new world of unfamiliar rules, systems, and power structures. Today, we're all immigrants in this emerging AI-mediated landscape, trying to navigate an environment where the "rules" of engagement are increasingly opaque and potentially adversarial. Anyone interested in buying the Brooklyn Bridge with me? I know a guy that is offering me an exclusive, inside deal... and we can pay with Klarna over a few payments.
The parallel extends further - just as those immigrants needed to develop new frameworks to thrive in their new world, we need new frameworks for digital well-being. We've moved beyond simple heuristics like "buy organic" for healthy living. The challenge is that there's no "certified organic" label for digital content, and the damage from manipulative content can far exceed that of a few pesticide-sprayed grapes.
I'm particularly intrigued by your concept of guardian angel agents. It reminds me of how religious institutions once provided moral guidance through periods of massive social change. Perhaps AI tools, properly designed and aligned with human values, could serve a similar role - helping us maintain our agency while navigating this new landscape. While we're at it, how about some AI generated hymns to go along with it?
The key seems to be combining:
Education about how AI tools actually work and influence us
Critical examination of our personal values and the legacy we want to build
Thoughtful deployment of AI tools that protect rather than exploit
Thank you as always for helping to fire the neurons!
Love your thoughts here Stephen! I'm feeling more and more motivated to work on the guardian angel. It seems the headwinds of unregulated tech and a bifurcation of trust will make adoption tricky — not the mention the impending arms race of AI agent persuasion. But damned if it's not an exciting mission!
Good post man. The social component can't be overstated. Social media and platforms are a means by which individuals obtain presence, affirmation, acknowledgement, and other passive and active forms of social validation. I think this is true for people who are mainly "lurkers" as well as those who are "influencers." Fact that these social relationships are in part imaginary, or at least projected and internalized, doesn't seem to mitigate their power in a society within which so much communication and social interaction is mediated. Mediated relationships have become as "real" as real life, and we practice the same types of social co-presence online as we would in real face to face encounters (albeit without same consequences). So the impact of social platforms is social/relational/psychological as it is also communication/distribution/content. Two axes by which to critique this mode of social interaction and production of social narratives.
The AI angle is interesting and I've thought a fair bit about it. To date AI has not leveraged social sharing; we use GPT privately (or for/at work). LLMs have a tough time with sentiment analysis, personalization, individual preference mapping, etc, so while interesting it's not there yet either as a mode of integrating AI into social platforms or for the guardian angel role you describe. (Social platforms e.g. Facebook might be rolling out AI avatars but think this is just proof of concept stuff as personalization and emotional realism are difficult frontiers for models trained on aggregated/generic linguistic content.) We should chat on LI about this - lack of social sharing w GPT, Claude, Gemini is worth fleshing out - what would they have to change to become acceptable sources, or for interactions with LLMs to be socially valid conversations?
Thanks for the comments Adrian! Thoughtful as usual :)
re. mediated relationships... I often think about Dunbars Number, a pre-digital take on our capacity for social connection. One way to think of the impact of social media is it's robbing our potential capacity for nurturing our community relationships. Virtual relationships are empty calories.
re. social + AI... it's coming. Meta is sitting on a massive pile of personalization insights, but have been constrained by the dynamics of the ad model, plus the meager regulatory constraints around the ad market. Embedding AI into the marketing tools to contextually deliver persuasive content will fly under the radar of current regulation. It's the low hanging fruit to boost their ad business and drive a much higher hit rate. Lots they can do without directly connecting AI to consumers.